Preparing your space
Your library and workspace are being arranged now.
Preparing your space
Your library and workspace are being arranged now.
Loading the business overview…
Explore the research to see how business evidence informs assumptions and valuation.
Start with the business evidence, then follow the assumptions through to scenarios and valuation.
Valuation and assumptionsAlphabet is a wide-moat compounder undergoing a transition from asset-light advertising to capital-intensive AI infrastructure, whose economics are primarily driven by the durability of Search advertising growth under AI substitution, the trajectory of capital expenditure through 2030, and the conversion of Google Cloud's backlog. Its principal competitive advantage is a combination of distribution across Search, Android, Chrome and YouTube, two decades of index and query data, and - uniquely among AI-era platforms - its own accelerator silicon, which from Q2 2026 it began selling into customers' own data centres rather than only renting access to; both breakup attempts against it have now failed in court. Its most important vulnerabilities are that no network effect protects Search from a change in the consumer interface, that capital expenditure guided at $195 to $205 billion for 2026 has taken quarterly free cash flow negative for the first time in company history, and that incremental returns on invested capital have fallen from roughly 30% to about 11%. Under the base case, revenue compounding at 11.3% to $1.3 trillion by 2035 with operating margin dipping to 33% on depreciation before recovering to 37%, and capital expenditure peaking near $250 billion in 2028 before converging toward depreciation, implies intrinsic value of approximately $300 per share versus a market price of $339.49. The investment therefore depends primarily on whether AI infrastructure spending settles at a permanently higher share of revenue or normalises toward historical levels - a question worth roughly $50 per share on which no participant in this industry has yet reached a steady state. The most important evidence to monitor is Search and other advertising revenue growth, Cloud revenue growth and backlog, the annual capital expenditure guide, depreciation as a percentage of revenue, and whether TPU system revenue is separately disclosed at scale.
| Field | Value |
|---|---|
| Company | Alphabet Inc. |
| Ticker / exchange | GOOGL / NASDAQ (Class A) |
| Current share price | $339.49 (2026-09-03) [S7]; $337.37 on 2026-09-02 [S14]; GOOG $334.31 [S13] |
| Market capitalization | ~$4.17 trillion [Est., 12,274.0M diluted shares [V] × $339.49]; reported as $4.12T [S7] |
| Enterprise value | ~$4.17 trillion [Est., roughly net-cash-neutral — see §12] |
| Industry | Digital advertising, cloud computing, AI infrastructure |
| Primary geographies | United States, EMEA, Asia-Pacific, Canada and Latin America [S12] |
| Investment classification | Compounder |
| Business quality | High |
| Competitive advantage | Strong |
| Financial quality | Strong (degrading on cash conversion) |
| Growth outlook | Strong |
| Balance-sheet risk | Low |
| Valuation attractiveness | Fair |
| Overall risk | Medium |
| Base-case intrinsic value | $300 |
| Bull-case intrinsic value | $480 |
| Bear-case intrinsic value | $165 |
| Probability-weighted value | $320.25 [Est.] |
| Upside/downside vs price | −5.7% |
| Expected annualized return, base case (3–5yr) | +4% to +9% [Est.] |
| Investment stance | Hold |
Sell-side is 58 buy and zero sell, with an average target of $428.07 and a low of $340 [S14] — meaning the lowest published target is above the current price and every one of them is above my base case.
Why it is attractive. The operating business is accelerating at a scale where that should be impossible. Q2 revenue of $119.8bn grew 24% — the twelfth consecutive quarter of double-digit growth [S3] — with operating income up 30% to $40.8bn and operating margin expanding two points to 34% [S1]. Google Cloud revenue rose 82% to $24.8bn with operating income nearly tripling to $8.8bn and backlog reaching $514bn [S1][S6]. And on 2 September the ad-tech case ended without a breakup [S8][S9][S10], removing the second of two existential legal threats within twelve months.
What the market may be misunderstanding — in both directions. The headline trailing P/E of roughly 17× [S7] is computed on net income containing $99bn of unrealized gains on SpaceX and Anthropic stakes [S5]. On operating earnings the multiple is closer to 30× [Est., §23]. That cuts against the bulls. Cutting the other way: unlike a pure AI capex story, Alphabet sells the capacity it builds — the $514bn Cloud backlog [S6] is contracted demand that Meta's comparable buildout does not have.
The most important value driver. Search and Other advertising, at $63.3bn a quarter growing 17% [S1] — roughly half of company revenue and the overwhelming majority of profit. A sustained three-point change in its growth rate is worth roughly $55 per share [Est., §17].
The largest risk. Capital intensity becoming permanent. FY2026 capex is guided to $195–205bn, raised from $180–190bn [S2], and the CFO says 2027 will "significantly increase" from there [S3]. Quarterly free cash flow turned negative for the first time in company history at −$5.9bn [S6]. The depreciation from roughly $450bn of 2026–28 capex lands in the P&L through 2032.
What must be true. That Search advertising holds double-digit growth through the transition to AI-mediated interfaces; that Cloud's $514bn backlog converts on schedule (management guides just over half within 24 months [S6]); and that capex converges toward depreciation before the depreciation overwhelms the margin. All three, not one.
This is the first Wide rating I have given in this research series, and it rests on four advantages that are individually strong and jointly very hard to attack.
Search, Chrome, Android, YouTube, Maps and Play give Alphabet default access to a majority of the connected world. Two courts have now declined to break that apart [S8][S10]. The Mehta remedies end exclusive default deals and require some search-data sharing [S8], which is a real cost — but the structure survives.
Two decades of crawling, query logs and click feedback. Replicable in principle, prohibitively expensive in practice. What weakens it: the Mehta requirement to share certain search data with qualified competitors [S8] is a deliberate, court-ordered erosion of exactly this advantage, and its magnitude is not yet knowable.
Alphabet designs TPUs and, from Q2 2026, sells TPU systems into customers' own data centres [S3]. This gives it a cost-per-FLOP advantage over competitors renting merchant silicon, and it converts an internal capability into a revenue line with the vast majority of contracted revenue expected in 2027 [S3].
Of everything in this report, this is the advantage I think is most underpriced. Every other participant in the AI buildout — Meta, the neoclouds, OpenAI — is a price-taker to NVIDIA. Alphabet is not.
$514bn of backlog [S6] with 82% growth [S1] and 35% segment margins [Est.]. Switching costs in cloud are genuine once workloads are deployed.
$11.1bn of quarterly advertising [S1] plus subscription revenue, with a creator supply chain nobody has replicated.
No network effect protects Search from an interface change. My use of Google does not improve yours. If the question-answering interface moves to an assistant that does not route through Google, distribution and index depth defend the franchise less than they appear to. That is the one mechanism that could take a Wide moat to Narrow inside a decade, and it is why §17's driver #1 is what it is.
| Test | Result |
|---|---|
| ROIC above WACC | Yes, substantially — see §11 |
| Rising market share | Yes in Cloud (82% growth); unknown in Search query share |
| Pricing power | Present but not disclosed — no volume/price split for Search [Unverified] |
| Operating margin | 34–35% [V][S1], expanded 7.6 points in three years |
| Incremental margin | FY2024→FY2025 revenue +$52.8bn produced operating income +$16.7bn — 32% [Est. from V] |
| Cloud backlog | $514bn [S6], roughly 1.04× annualised revenue |
| Customer concentration | None material |
The incremental-margin row deserves a note. At 32% [Est.], incremental margin roughly equals the average margin — meaning Alphabet is growing without operating leverage at the consolidated level, because Cloud's lower margins and rising depreciation offset Services' leverage. That is a fair description of a business in mid-transition, and it is not the picture of accelerating leverage that the headline margin expansion suggests.
| Risk | Type | Prob. | Financial impact | Horizon | Leading indicator | Mitigating factors | Priced in? |
|---|---|---|---|---|---|---|---|
| AI capex fails to earn its cost of capital | Business | 35% | −$70 to −$110/share | 3–6yr | Incremental ROIC; FCF recovery timing | $514bn backlog; capex is discretionary | Partially |
| Search substitution by AI assistants | Business | 30% | −$80 to −$130/share | 4–10yr | Search revenue growth; assistant adoption | Gemini, AI Mode, distribution | Poorly |
| Depreciation compresses margin structurally | Business | 65% | −$30 to −$50/share | 2–4yr | D&A as % of revenue | Cloud mix improving | No |
| Advertising recession | Business | 25% in any 3yr window | −$45/share peak-to-trough | Any | SMB ad surveys; consumer confidence | No customer concentration; capex flexible | No |
| Sealed Brinkema remedies prove costly | Regulatory | 30% | −$12 to −$25/share | 0–2yr | The opinion, unsealed in 14 days | No breakup; behavioural only | Partially |
| Mehta remedies on appeal go against Alphabet | Regulatory | 25% | −$15 to −$30/share | 1–3yr | Appellate calendar | Already operating under them | Partially |
| Equity-stake marks reverse | Financial | 45% | Reported earnings only | Any | SpaceX and Anthropic valuations | Non-cash; already volatile | Partially |
| Rising leverage constrains flexibility | Financial | 25% | Option value loss | 2–4yr | Debt/EBITDA; issuance size | Roughly net-cash today | No |
| EU ad-tech proceedings | Regulatory | 40% | −$8 to −$15/share | 1–3yr | Publisher feedback process [S14] | US outcome sets precedent | No |
| Addiction and product-liability litigation | Legal | 35% | −$5 to −$15/share | 2–5yr | Docket; the Meta settlement precedent | Diversified; deep pockets | No |
| Multiple compression on capital intensity | Valuation | 45% | −$50/share | 1–3yr | Peer multiples; FCF yield | FCF recovery | Partially |
| Balance-sheet distress | Financial | <2% | Immaterial | — | — | ~$100bn liquidity | Yes |
Distinguishing the three types. Business risk is moderate and concentrated in two questions. Financial risk is very low. Valuation risk is the dominant one — at roughly 30× operating earnings with zero current free cash flow, the price embeds a recovery that has not yet begun.
| Assumption | Value | Basis |
|---|---|---|
| Risk-free rate | 4.3% | [Unverified] |
| Beta | 1.05 | [Est.] — Alphabet's long trading history makes this reasonably measurable |
| Equity risk premium | 5.0% | [Est.] |
| Cost of equity | 9.55% | Derived |
| Cost of debt (pre-tax) | 5.0% | [Est., Unverified — maturity schedule not read] |
| After-tax cost of debt | 4.15% | Derived |
| Debt weight | 2.4% | $100.16bn [V] against ~$4.17tn equity |
| WACC | 9.5% | Derived, rounded |
| Terminal growth | 3.0% | [Est.] |
| Tax rate | 17% | [Est., Unverified — Alphabet's effective rate has run in the mid-to-high teens] |
| Diluted share count | 12,274.0M | [V], six months to 2026-06-30 |
| Share count assumption | Held flat despite a demonstrated 7.1% reduction FY2022–FY2025 [V] | Deliberately conservative |
| Stock-based compensation | Fully expensed. Verified operating income of $80.47bn [V] is GAAP and already deducts SBC of roughly $14bn for the half [Est.]. | Deliberate |
| Equity investments | Credited at $150bn [Est., Unverified] in the equity bridge | Not in the operating forecast |
Two treatment notes for comparability:
SBC. Not stored by the platform and estimated at ~$26bn a year [Est.]. It is already inside the verified operating income I use. A valuation built on an "adjusted" operating figure that adds it back would be roughly 5 points of margin more generous.
Equity stakes. I add $150bn [Est.] to the equity bridge because SpaceX and Anthropic holdings generate none of the modelled operating cash flows. A valuation ignoring them understates by about $12/share; one marking them to a bull-case SpaceX overstates by more.
Growth vs reinvestment vs incremental ROIC — consistent, and deliberately unfavourable. I forecast revenue compounding at 11.3% to 2035 while assuming capex stays at $195–250bn through 2030 and settles near 16% of revenue. I am not assuming growth arrives while reinvestment falls. The consequence is negative free cash flow in 2026–27 and a recovery only from 2028. Anyone reaching my revenue path with capex normalising to 12% of revenue is claiming rising incremental returns and owes an explanation of what produces them — particularly given §11 shows incremental ROIC has fallen from ~30% to ~11%.
Pricing vs volume vs market share — under-evidenced and I want to flag it. I assume Search revenue growth of low-to-mid teens decaying toward high single digits, without separating volume from price. Alphabet does not disclose the split, so I cannot verify that my growth assumption is volume-supported rather than ad-load-supported. If it is the latter, it has a ceiling I have not modelled. This is a genuine hole and it exists because of a disclosure gap, not an analytical choice.
Margins vs competitive intensity vs required investment — strained, and I resolve it in the market's favour rather than mine. I hold terminal operating margin at 37%, above today's 34–35%, while assuming Cloud (lower margin than Services) grows to a third of revenue and depreciation rises steeply. Those pull against each other. My defence is that Cloud margins are still expanding rapidly toward Services levels — 35.5% segment margin already [Est. from S1] — so the mix drag is smaller than it looks. But if Cloud margins plateau in the low 30s, 37% terminal is too high.
Capex vs depreciation vs free cash flow — this is where the model is tightest and it is the crux. I forecast capex falling from a peak near $250bn (2028) to roughly $200bn by 2035, against D&A rising to $195bn. They converge, which is what produces the free cash flow recovery. If capex does not converge — if AI infrastructure requires permanent replacement at the deployment rate — free cash flow stays structurally depressed and my base case is 25–30% too high. I cannot resolve this from any disclosure and neither can anyone else.
Terminal growth vs terminal reinvestment vs terminal ROIC — consistent. 3.0% growth with capex ≈ D&A implies terminal ROIC well above WACC, which matches Alphabet's history even after the decline in §11.
All figures $bn, calendar years. GAAP operating margin — SBC expensed.
| Year | Revenue | Op. margin | EBIT | NOPAT (17%) | D&A | Capex | FCF | PV @9.5% |
|---|---|---|---|---|---|---|---|---|
| 2026E | 495 | 34% | 168 | 139 | 45 | 200 | (16) | (14) |
| 2027E | 585 | 33% | 193 | 160 | 75 | 240 | (5) | (4) |
| 2028E | 675 | 33% | 223 | 185 | 105 | 250 | 40 | 30 |
| 2029E | 755 | 34% | 257 | 213 | 130 | 240 | 103 | 72 |
| 2030E | 840 | 35% | 294 | 244 | 150 | 220 | 174 | 111 |
| 2031E | 920 | 36% | 331 | 275 | 165 | 205 | 235 | 136 |
| 2032E | 1,000 | 36% | 360 | 299 | 178 | 200 | 277 | 147 |
| 2033E | 1,080 | 36.5% | 394 | 327 | 190 | 200 | 317 | 153 |
| 2034E | 1,190 | 37% | 440 | 365 | 205 | 205 | 365 | 161 |
| 2035E | 1,300 | 37% | 481 | 399 | 215 | 210 | 404 | 163 |
Base case = $300.
Terminal value is 73% of enterprise value. That is high, and it is unavoidable when free cash flow is negative for two of the next three years. It also means this valuation is substantially a statement about 2035, not about 2027.
Note the forecast does not credit the buyback. Share count is held flat despite a 7.1% reduction over three years [V]. Continuing at that pace adds roughly $18–22 per share over five years [Est.] that is not in the $300.
Detailed in §22. The current price requires roughly 19% more present value than my base case — achievable through any single one of: 12.9% revenue CAGR, 39% terminal margin, 13% terminal capex intensity, or an 8.9% WACC.
| Multiple | Alphabet @ $339.49 | Comment |
|---|---|---|
| P/E (TTM) | ~17× [S7] | On net income containing ~$99bn of unrealized marks — do not use |
| P/E on FY2026E operating earnings (~$11.30) | ~30× [Est.] | The economically meaningful figure |
| EV / FY2026E revenue | ~8.4× [Est.] | |
| EV / FY2026E EBIT ($168bn) | ~21× [Est.] | |
| EV / FY2026E EBITDA (~$213bn) | ~19.6× [Est.] | |
| Price / FCF (FY2026E ~$0) | n/m | |
| FCF yield | ~0% | Against a 4.3% risk-free rate |
| Price / FY2029E FCF ($103bn) | ~40× | |
| Dividend yield | 0.25% [S7] | Token |
The second and last rows are the honest summary. Alphabet at 17× trailing earnings looks like one of the cheapest megacaps in the market. Alphabet at 30× operating earnings with a zero free cash flow yield is a full valuation for a high-quality asset. Both statements describe the same security, and the first is an artefact.
Should Alphabet trade at a premium? Yes — 24% revenue growth at $4tn scale, a wide moat, unique silicon integration, and two breakup threats resolved. A premium to the market, certainly. Thirty times operating earnings with no current cash return is where quality meets a full price, which is what "fair" means.
Sum-of-the-parts [Est., illustrative]:
| Part | Method | Value |
|---|---|---|
| Google Services | 20× normalized after-tax Services earnings of ~$135bn [Est.] | ~$2,700bn |
| Google Cloud | 12× 2028E Cloud revenue of ~$150bn [Est.] | ~$1,800bn |
| Other Bets | Capitalized loss stream | −$60bn |
| Equity investments | Marked | +$150bn |
| Net cash | ~$0bn | |
| Implied | ~$4,590bn = $374/share |
I do not believe that number and include it to show why. It values Services and Cloud on normalized earnings and revenue multiples that assume the capex problem away — the capital being spent appears nowhere. The gap between the $374 SOTP and the $300 DCF is precisely the capital intensity question, priced. The market at $339.49 sits between them, which is a coherent place to sit.
Earnings power value (no growth, FY2026E maintenance-capex owner earnings of $139bn capitalized at 9.5%): $1,463bn = $119/share. So roughly 65% of the current price is growth — comparable to SpaceX (65%) and far below Palantir (94%) in this series.
Each scenario is a causal chain: trigger → operating mechanism → financial consequence → valuation consequence.
Trigger: AI assistants absorb a growing share of commercial queries faster than Gemini and AI Mode can defend, while the industry-wide capex cycle decelerates in 2028.
Operating mechanism: Consumers increasingly begin purchase research inside assistants — some Google's, many not. Search query volume growth stalls, and Alphabet initially masks it by raising ad load and improving conversion prediction, so revenue growth holds up for two years before decelerating sharply. Meanwhile the Brinkema behavioural remedies [S9][S10] and the Mehta search-data-sharing requirement [S8] modestly advantage competitors in ad tech and search. Cloud growth halves as the enterprise AI budget cycle normalises and the $514bn backlog converts more slowly than the 24-month guidance [S6]. Depreciation from $450bn of 2026–28 capex arrives in full against decelerating revenue; because roughly 60% of that spend is servers on short lives [S4], the charge is front-loaded. Alphabet cuts capex hard, which stabilises cash flow but confirms that part of the buildout was stranded.
Financial consequence: revenue growth falls to mid-single digits by 2030, reaching roughly $800bn by 2035. Operating margin compresses to 28–30% as depreciation outruns the Cloud mix benefit. Free cash flow recovers only to $120bn.
Valuation consequence: terminal FCF around $170bn, WACC 10.5% on a de-rated growth story, terminal growth 2.5%. Intrinsic value: $165 — a 51% decline, and still a business earning well above its cost of capital.
Trigger: Alphabet successfully becomes the assistant as well as the search engine, and AI infrastructure spending proves to be a multi-year investment cycle rather than a permanent step-change in capital intensity.
Operating mechanism: Gemini and AI Mode retain commercial queries within Google's surfaces, so Search revenue growth decelerates gradually from 17% toward high single digits by 2031 rather than breaking. Cloud continues taking share, decelerating from 82% toward 25% by 2030 as the base grows, with the $514bn backlog converting broadly on the guided schedule [S6] and segment margins holding in the mid-30s. TPU external system sales become a real but secondary revenue line from 2027 [S3]. Capex peaks near $250bn in 2028 and plateaus; depreciation rises to meet it, converging by the early 2030s. Third-party capacity use, and its associated margin pressure [S3], proves temporary.
Financial consequence: revenue $755bn (2029) and $1,300bn (2035); operating margin dipping to 33% in 2027–28 on depreciation, then recovering to 37%; free cash flow negative in 2026, roughly breakeven in 2027, then $103bn (2029) rising to $404bn (2035).
Valuation consequence: WACC 9.5%, terminal growth 3.0%. Intrinsic value: $300, excluding roughly $18–22 of buyback accretion not credited.
Trigger: Alphabet's vertical integration — its own silicon, its own models, its own distribution — proves to be a structural cost advantage that compounds as AI workloads scale.
Operating mechanism: TPUs give Alphabet a materially lower cost per unit of AI compute than any competitor renting merchant silicon, and selling TPU systems into customers' data centres [S3] turns that advantage into a large, high-margin third business alongside Services and Cloud. Cloud sustains growth above 50% into 2029 as enterprises consolidate onto the cheapest capable stack, and the backlog grows faster than it converts. Gemini becomes a genuine consumer assistant franchise, so the interface shift happens within Google rather than away from it, and AI-mediated commercial queries monetise at higher rates than blue links because intent is clearer. Because Alphabet builds its own accelerators, capex per unit of delivered compute falls even as absolute capex stays high — so capacity grows faster than spending. Capex peaks in 2028 and declines in absolute terms from 2030.
Financial consequence: revenue compounds at 14% to roughly $1.6tn by 2035; operating margin reaches 40% as silicon economics and Cloud scale compound; free cash flow exceeds $550bn. The equity re-rates as duration risk falls and the market treats the capex as a completed investment rather than an open-ended commitment, compressing the discount rate toward 8.75%.
Valuation consequence: WACC 8.75%, terminal growth 3.25%. Intrinsic value: $480 — above the 52-week high of $408.61 [S7] and near the $515 high published target [S14].
| Scenario | Value | Probability | Weighted |
|---|---|---|---|
| Bear | $165 | 0.25 | $41.25 |
| Base | $300 | 0.45 | $135.00 |
| Bull | $480 | 0.30 | $144.00 |
| Total | 1.00 | $320.25 |
These sum to 1.00 and I am satisfied they should. Unlike the SpaceX and IREN records in this series, Alphabet has no unvaluable distress branch — the balance sheet removes it — and the three scenarios span the outcome space: the interface shift helps Alphabet, hurts it, or is navigated.
Versus $339.49, the weighted value implies −5.7%.
Note the shape. The distribution is roughly symmetric in probability but positively skewed in value, and the bull contributes 45% of the weighted value from 30% probability. If you believe the bull case is 45% likely rather than 30%, the weighted value rises to about $347 and the stance becomes a marginal Buy. That is a smaller swing than in most reports in this series, which is a way of saying this conclusion is unusually robust to the probability judgement.
34–35% operating margin at $495bn of revenue [V][S1], 24% growth at $4tn of scale, no customer concentration, and a Cloud business that reached hyperscaler margins in a single year. Marked down one point because capital intensity has gone from 12% to 40% of revenue in six years [V][Est.] and the marginal economics are no longer asset-light.
Wide moat, and the only one I have rated Wide in this series: distribution across Search, Android, Chrome and YouTube; two decades of index and query data; a Cloud backlog of $514bn; and — uniquely — its own accelerator silicon, now sold externally [S3]. Two courts have declined to break it apart [S8][S10]. Not 10 because no network effect protects Search from an interface change.
An AI response far better than the market expected two years ago, a genuinely differentiated silicon programme, honest cost disclosure, and consistent buybacks reducing shares 7.1% in three years [V]. Marked down for repeated upward capex revisions with no disclosed return framework, and for dual-class governance.
Roughly neutral net debt, ~$100bn of liquidity, and a business that stays comfortably free-cash-flow positive in any stress case I can construct. Marked down only because long-term debt doubled to $98.16bn in six months [V].
24% consolidated growth, 82% in Cloud, a $514bn backlog, and a new silicon revenue line arriving in 2027.
About 30× operating earnings [Est.], zero free cash flow yield, and a base case 12% below the price. Cheap on the headline P/E only because that number is an artefact.
Bear −51%, base −12%, bull +41%, with a probability-weighted −5.7%. The mildest bear case in this series, and a positively skewed but modest distribution.
1. Is this a good business? It is an outstanding one — the highest-quality business in this research series on moat width, competitive position and financial strength. The qualification is that its returns on incremental capital have fallen from roughly 30% to about 11% (§11).
2. Stronger or weaker? Operationally stronger and economically more capital-hungry. Revenue accelerating to 24%, Cloud margins transformed, two breakup threats resolved. Against that: free cash flow at roughly zero, debt doubled in six months, and incremental ROIC near the cost of capital.
3. Durable competitive advantage? Yes — wide. Distribution, data, YouTube, Cloud scale, and vertical silicon integration no competitor matches. The durability question is the interface, not the technology.
4. Three most important value drivers? (i) Search advertising growth durability under AI substitution; (ii) the capital expenditure trajectory through 2030; (iii) Cloud growth and backlog conversion.
5. Three largest risks? (i) AI assistants displacing commercial search queries; (ii) AI capex failing to earn its cost of capital; (iii) depreciation from 2026–28 capex compressing margin faster than revenue absorbs it.
6. Most important dependency? Power and data-centre construction capacity. Management is buying third-party capacity as a bridge and accepting margin pressure to do it [S3] — an explicit statement that the physical constraint, not demand or capital, is binding.
7. What is the market pricing in? Roughly my base case plus about 19% — achievable through any one of a 12.9% revenue CAGR, a 39% terminal margin, 13% terminal capex intensity, or an 8.9% WACC. None of those is unreasonable, which is why the disagreement is narrow.
8. Where could the market be wrong? On the trailing P/E, which is an artefact of $99bn in unrealized equity marks. And on under-pricing TPU external sales, which convert a cost advantage into a market from 2027 [S3].
9. Fair intrinsic value? $300 base case; $320.25 probability-weighted; an honest range of $165 to $480.
10. Margin of safety today? None. The price is 13% above base case and 6% above the weighted value — agreement rather than margin.
11. Reasonable annualized return over 3–5 years? +6.4% probability-weighted. Bear −8.2%, base +6.8%, bull +17.9%. Note that fundamentals alone contribute about +16% a year in the base case; the shortfall is assumed multiple compression from 30× toward 21×.
12. Single development that would most strengthen the thesis? Disclosed TPU system revenue reaching $15bn or more in a fiscal year — proof that the silicon advantage is a market rather than an internal saving.
13. Single development that would most invalidate it? Search and other advertising growth falling below 8% for two consecutive quarters absent a macro cause — the signature of the interface shift arriving faster than Gemini can absorb it.
14. At what price does it become clearly attractive? $255 — a 15% discount to base case, and a level at which my terminal capex assumption could be two points too high and an investor would still not lose money. Note that is above the 52-week low of $224.79 [S7].
15. What price implies excessive optimism? $480 — the bull case, requiring the silicon advantage to compound into a third major business and Gemini to own the assistant interface.
Alphabet is a wide-moat compounder undergoing a transition from asset-light advertising to capital-intensive AI infrastructure, whose economics are primarily driven by the durability of Search advertising growth under AI substitution, the trajectory of capital expenditure through 2030, and the conversion of Google Cloud's backlog. Its principal competitive advantage is a combination of distribution across Search, Android, Chrome and YouTube, two decades of index and query data, and — uniquely among AI-era platforms — its own accelerator silicon, which from Q2 2026 it began selling into customers' own data centres rather than only renting access to; both breakup attempts against it have now failed in court. Its most important vulnerabilities are that no network effect protects Search from a change in the consumer interface, that capital expenditure guided at $195 to $205 billion for 2026 has taken quarterly free cash flow negative for the first time in company history, and that incremental returns on invested capital have fallen from roughly 30% to about 11%. Under the base case, revenue compounding at 11.3% to $1.3 trillion by 2035 with operating margin dipping to 33% on depreciation before recovering to 37%, and capital expenditure peaking near $250 billion in 2028 before converging toward depreciation, implies intrinsic value of approximately $300 per share versus a market price of $339.49. The investment therefore depends primarily on whether AI infrastructure spending settles at a permanently higher share of revenue or normalises toward historical levels — a question worth roughly $50 per share on which no participant in this industry has yet reached a steady state. The most important evidence to monitor is Search and other advertising revenue growth, Cloud revenue growth and backlog, the annual capital expenditure guide, depreciation as a percentage of revenue, and whether TPU system revenue is separately disclosed at scale.
Most of the above are readable in the Q2 2026 Form 10-Q, the FY2025 Form 10-K and the 2026 proxy. This report should be re-run against those documents before any capital is committed, re-run again when the Brinkema opinion is unsealed in mid-September, and re-run after Q3 2026 results on 27 October [S14].
Terminal capital intensity is the weakest link, worth roughly $50 per share and the largest gap between this estimate and the market's; no participant in this industry has reached a steady state, and Alphabet's own capex-to-revenue has moved from 12% to roughly 40% in six years. The discount rate is a second methodological lever: an 8.5% WACC would give $384, above the current price, so two judgement calls account for essentially the whole gap between the $300 base case and the market. Depreciation and amortization and stock-based compensation are both explicitly not stored by the platform and were not read from the 10-Q; the D&A ramp is central to the entire margin forecast, and server depreciation lives are the highest-leverage accounting estimate, worth roughly $18 per share for a one-year change. Also unresolved: the volume versus price split in Search advertising, which Alphabet does not disclose, so growth cannot be verified as volume-supported rather than ad-load-supported; segment operating income outside Cloud; the Brinkema ad-tech remedies themselves, since the full opinion remains under seal for fourteen days and only the absence of a breakup is known; marketable securities and the carrying value of the SpaceX and Anthropic equity stakes, credited at an estimated $150bn; the effective tax rate, assumed at 17%; the debt maturity schedule; executive compensation and Class B voting mechanics; and geographic revenue mix and competitor financials. Note also that H1 2026 net income of $174.77bn exceeds operating income of $80.47bn because of roughly $99bn of unrealized equity gains, so the widely quoted trailing P/E of about 17 times is an artefact and the operating multiple is closer to 30 times.
These model scenarios depend on their inputs. Review the assumptions and supporting evidence alongside the implied values.
earnings multiple v2 · as of 2026-09-12
Sign in for the price chart, where every input traces back to a filing, an adjustable scenario sandbox, and companies that move with GOOGL.
Create a free account →